Objective. Universal childhood immunization against varicella in the United States, first recommended in 1995, was predicted to lead to significant decreases in varicella-related hospitalization rates and corresponding charges. Previous studies have not found such effects. We studied trends in varicella-related hospitalization rates and associated charges before and after introduction of varicella vaccine.
Methods. We examined hospitalization and charge data from the Nationwide Inpatient Sample for the years 1993–2001, representative of national hospitalization patterns for children and adults. We derived weighted estimates of population-adjusted, varicella-related hospitalization rates and inflation-adjusted, varicella-related hospital charges.
Results. The annual varicella-related hospitalization rate exceeded 0.5 hospitalizations per 10 000 US population from 1993 to 1995, declined to 0.26 per 10 000 by 1999, and again halved to 0.13 per 10 000 by 2001. Hospitalization rates declined most substantially among individuals primarily targeted for vaccination (0- to 4-year-old children) but decreased among youths aged 5 to 19 years and among adults as well. Concomitantly, varicella-related hospital charges declined from $161.1 million (95% confidence interval: $130.5 million–$191.8 million) in 1993 to $66.3 million (95% confidence interval: $50.9 million–$81.7 million) in 2001 (all 2001 $US). Among expected primary payers, inflation-adjusted declines in varicella-related hospital discharges–related charges accrued to Medicaid, private insurance, and “other” payers (including uninsured and self-pay) but not to Medicare.
Conclusions. This national analysis indicates a clinically and statistically significant reduction in varicella-related hospitalizations for children and adults associated with childhood varicella immunization in the United States and a corresponding significant decrease in hospital charges.
In May 1995, live attenuated varicella vaccine was recommended for universal administration to children ≥12 months of age by the American Academy of Pediatrics.1 At the time, existing epidemiologic and economic models anticipated that national immunization against varicella would be associated with prompt decreases in the number of hospitalizations and costs attributable to severe varicella disease.2,3
Subsequent analyses, however, have not indicated significant effects of varicella vaccination on varicella-related hospitalization. Three studies of hospitalization patterns found no significant difference in rates of varicella-related hospitalizations before and after the vaccine was introduced, using local4 and national5,6 data. A separate investigation of varicella in 3 communities with active disease surveillance from 1995 to 2000 found a significant decline in the number of varicella cases with increasing community immunization rates but no significant concurrent trend in varicella-related hospitalizations.7
These previous studies were hindered by sample sizes that either were insufficient to detect small differences in hospitalization rates4,7 or required the investigators to pool data from multiple years, thereby losing the opportunity to examine year-to-year trends.5,6 Moreover, none of the foregoing studies measured medical expenditures associated with varicella-related hospitalizations.
In this report, we present an analysis of data from the Nationwide Inpatient Sample (NIS), a nationally representative annual sample of discharges from nonfederal, short-term, general, and other specialty hospitals in the United States, including children's hospitals.8 The NIS includes International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes and total charges for each of >6 million discharges annually, which permits greater power to examine year-to-year patterns of varicella-related hospitalizations than in previous studies.5,6
The NIS is a component of the Healthcare Cost and Utilization Project (HCUP), a set of health care databases developed through partnership among federal and state governments and health care institutions, sponsored by the Agency for Healthcare Research and Quality.8 The NIS is the largest publicly available all-payer inpatient database in the United States.
The NIS contains deidentified, patient-level clinical and resource use data included in a typical discharge abstract. For each year, these data reflect hospital stays from between 800 and 1000 institutions sampled to approximate a 20% stratified sample of nonfederal community hospitals, including public hospitals, children's hospitals, and academic medical centers but excluding long-term hospitals, psychiatric hospitals, and chemical dependency treatment facilities. For NIS sampling purposes, the universe of US hospitals is divided into strata using 5 hospital characteristics: US region, urban/rural location, ownership/control, bed size, and teaching status.
The study period included the years 1993 through 2001. Starting with 1993, data permitted analyses of 3 years before the first full year of varicella vaccination efforts in 1996, and 1993 also marked the first year that the proportion of states that participated in HCUP exceeded one third (17 states). By 2001, the most recent year of NIS data available, 33 states were participating in HCUP.
For each year of NIS data during the study period, discharge-level weights were provided to permit calculation of national estimates of hospitalization rates standardized to the concurrent national population.8 We used US Census data9 for each of the years in the study period to obtain rate denominators for the overall population and for specific age groups of interest in varicella analyses: 0 to 4 years, 5 to 9 years, 10 to 19 years, and ≥20 years. To interpret the analyses of discharge-related hospital charges in constant dollars, we standardized all hospital charges to 2001 US dollars using the mean Consumer Price Index for each year of data.10
Identification of Varicella-Related Hospital Discharges
Consistent with methods described in previous studies of varicella-related hospitalizations,5,6,11 we searched for a varicella diagnosis (ICD-9-CM codes 052.xx) in any of 15 diagnostic positions included for each discharge in the NIS. We excluded hospitalizations with coexistent herpes zoster codes (053.xx) from analysis as varicella-related discharges, because the vaccine is not primarily indicated for prevention of herpes zoster.
We defined varicella-related hospital discharges (VRHDs) as those with 1) varicella as the principal discharge code (052–052.9); or 2) postvaricella encephalitis (052.0) or varicella pneumonitis (052.1) in any diagnostic position; or 3) varicella as any discharge code for a person with an immunocompromising condition likely to necessitate hospitalization for varicella (eg, human immunodeficiency virus/acquired immune deficiency syndrome, severe defect of T-cell immunity, organ recipient, malignancy, recipient of chemotherapy); or 4) varicella as any discharge code in a person with a condition that may increase the risk of varicella disease because of treatment of the condition (eg, rheumatoid arthritis, asthma, diabetes); or 5) a code for a well-described potential complication of varicella12–14 as the principal diagnosis and a varicella code in any subsequent diagnostic field. ICD-9 codes that correspond to immunocompromised conditions and to well-described potential complications of varicella are available from the authors on request. All discharges that were not categorized as varicella-related using these criteria were categorized as non–VRHDs.
National Childhood Immunization Rates Against Varicella
For purposes of comparing trends in rates of hospitalization with trends in immunization against varicella, we used data from the National Immunization Survey for the years 1996–2001.15–20 The National Immunization Survey is an annual, ongoing, nationally representative household survey that yields national estimates of immunization coverage among US preschool children aged 19 to 35 months. Varicella immunization rates were first measured in 1996.
Analyses were designed to test the study hypothesis that increasing community immunity to varicella as a result of national vaccination efforts is associated with decreased morbidity attributable to severe varicella illness, as measured by varicella-related hospitalizations and charges. The primary outcomes were annual population-adjusted rate of VRHDs (per 10 000 population) and inflation-adjusted total charges attributed to varicella-related hospitalizations. Secondary outcomes included proportion of varicella-related hospitalizations with well-known complications of varicella and distribution of total varicella-related hospitalization charges by expected primary payer.
The robustness of national estimates based on discharge weights in the NIS requires a minimum of 70 unweighted observations in each cell. Therefore, we were not able to examine rates of in-hospital mortality for varicella-related hospitalizations or the hospitalizations and charges for children who were younger than 1 year separate from other children in the 0 to 4 age group.
We conducted all analyses using SAS statistical software, version 8.2 (SAS Institute, Cary, NC), into which we incorporated sampling weights to account for the complex stratified sampling of hospitals that compose the NIS. Variance estimates with which we derived 95% confidence intervals (CIs) were also based on NIS sampling weights. Except for the sample description, all results are presented using weighted values.
The study was funded internally, and all analyses were conducted by the authors. The authors had no financial interest in the outcome. The study was exempt from human subjects review as an analysis of deidentified secondary data.
NIS samples for the years 1993–2001 included data for 6.4 million to 7.5 million discharges annually, representative of between 34.6 million and 37.2 million discharges nationally each year. The number of unweighted annual VRHDs ranged from a minimum of 740 to a maximum of 3270.
Rates of Varicella-Related Hospitalization
Population-adjusted rates of VRHDs in the United States declined significantly from 1993 to 2001 (Fig 1). Whereas the rate exceeded 0.5 VRHDs per 10 000 population from 1993 through 1995, the rate declined to 0.26 per 10 000 by 1999 and again halved to 0.13 per 10 000 by 2001. These declines all were statistically significant, as indicated by 95% CIs that do not overlap. The decreasing rate of VRHDs contrasts with the increasing rate of immunization against varicella among children aged 19 to 35 months during the same time period (Fig 1).
The decline in VRHDs was evident in population-adjusted hospitalization patterns across all age groups that we examined. Population-adjusted VRHD rates declined most substantially for the youngest group but for older groups as well (Fig 2). The more pronounced declines in the youngest age group led to a reversal in the relative burden of VRHDs among the youngest and oldest age groups over the course of the study period (Table 1). In 1993, a statistically significantly greater proportion of VRHDs occurred in the 0- to 4-year age group (41.3%) than among individuals ≥20 years of age (32.6%). By the end of the study period, that pattern had reversed such that 0- to 4-year-olds accounted for 27.8% of VRHDs and individuals ≥20 years of age accounted for 46.1% of VRHDs in 2001.
Complications of Severe Varicella Disease
The proportion of VRHDs with well-recognized complications (eg, pneumonitis, encephalitis, cellulitis, fasciitis) changed comparatively little from 1993 to 2001. The proportion of VRHDs with complications ranged from 51% to 54% from 1993 to 1995, then ranged from 39% to 54% from 1996 to 2001 without a consistent or statistically significant year-to-year trend (data not shown).
Hospital Charges Corresponding to VRHDs
Annual hospital charges for all diagnoses (including VRHDs) increased from $406.7 billion (95% CI: $383.9 billion–$429.2 billion) in 1993 to $548.8 billion (95% CI: $520.9 billion–$576.7 billion) in 2001 (all 2001 $US). During the same period of time, annual hospital charges for VRHD declined from $161.1 million (95% CI: $130.5 million–$191.8 million) to $66.3 million (95% CI: $50.9 million–$81.7 million), a decrease of almost $100 million annually. In other words, VRHD charges constituted 0.04% of all annual hospitalization charges in 1993, compared with 0.01% by 2001.
Among expected primary payers, constant-dollar declines in VRHD-related charges from 1993 to 2001 accrued to Medicaid, private insurance, and “other” payers (including uninsured and self-pay) but not to Medicare (Fig 3). Medicaid VRHD-related charges fell from nearly $75 million in 1993 to <$20 million by 2001. Private-payer VRHD-related charges declined from ∼$61 million to ∼$33 million during the same time period, whereas VRHD-related charges for “other” payers decreased from ∼$14 million to just over $4 million. Medicare VRHD-related charges ranged from ∼$10 million to ∼$20 million annually, depending on the study year, and were equivalent in 1993 and 2001.
This is the first analysis to establish a clinically and statistically significant decrease in population-adjusted rates of varicella-related hospitalization paired with the rise of childhood immunization rates against varicella in the United States. These findings support the hypothesis that national varicella vaccination efforts are associated with decreased rates of severe varicella disease. Earlier investigators using smaller single-hospital or national discharge data sets4–6 did not establish this association and attributed their findings to slow adoption by physicians of recommendations for the vaccine.21 However, data in this study indicate significantly decreased hospitalization rates by 1999 (the time period concurrent with earlier studies), and analysis through 2001 revealed even further rate declines.
Varicella-related hospitalization rates decreased across all age groups, concordant with the marked decline in cases of varicella illness reported in 3 well-immunized communities under active surveillance.7 This pattern is consistent with an epidemiologic model of a national varicella immunization program2 in which targeting the reservoir age group for the virus (0- to 4-year-olds) leads to an apparent herd immunity benefit for older individuals.
Economic Benefits of Varicella Immunization
Another salient finding from this analysis is the evident economic benefit of varicella immunization through decreased rates of varicella-related hospitalizations. A prelicensure model of the effects of universal childhood varicella immunization predicted a mean annual savings of ∼$35 to $40 million (converted to 2001 $US) in hospitalization costs, discounted annually at 5% over a 30-year immunization program.3 Converting these cost estimates to charges to be comparable with hospital charges in our analysis (assuming a typical ratio of hospital costs to charges of 0.5:1), the magnitude of predicted savings ($70–$80 million) is comparable to our estimate of almost $100 million saved annually. This is despite that the national varicella immunization rate among 19- to 35-month-olds in 2001 (76%20) was lower than the 90% immunization rate modeled3 and that hospitalization rates during the study period had decreased by ∼75% overall, not 94% as modeled.3
Nonetheless, changes over time in the patterns of VRHD-related hospitalization charges for different payers suggests that perspectives on the economic benefits of childhood varicella immunization may vary by payer. This concept is a departure from the customary approach to health economic analyses, in which investigators choose either a uniform payer perspective (direct medical costs alone) or a societal perspective (direct and indirect medical costs).22 In the case of varicella vaccine and other recommended childhood immunizations, payers may differ in their perspectives because some bear the purchase cost of varicella doses whereas others do not. Specifically, the federal Vaccines for Children (VFC) program (administered through the Centers for Disease Control and Prevention) and additional state immunization programs fund the purchase of ∼60% of all varicella vaccine doses administered in the United States annually. VFC-purchased doses are administered to children on Medicaid, children without insurance, children from Alaskan Native and American Indian backgrounds, and in some cases children whose private insurance plans do not cover the vaccine (the “underinsured”) if they receive the vaccine at a federally qualified health center or rural health center.23,24
Therefore, although VFC rather than Medicaid is billed for the purchase of varicella vaccine doses for Medicaid-enrolled children, Medicaid has reaped substantial benefits in terms of reduced numbers of varicella-related hospitalizations in the vaccine era. This is important because although both VFC and Medicaid are public programs, they are financed differently: VFC is supported solely through federal dollars, whereas states share the fiscal burden for Medicaid with the federal government. In essence, the federal government's investment in varicella vaccine for children on Medicaid has benefited both the federal government and the states, which may serve as an incentive to states to continue to invest in their immunization programs.
In contrast, although private payers have benefited from decreased VRHD-related hospitalization rates, they have also incurred the costs of reimbursement to providers for varicella vaccine purchase and administration. This is the cost-benefit trade-off faced by private payers that are not bound by state insurance mandates to cover recommended childhood immunizations: will the incremental benefits of an immunization offset the costs? Recent estimates suggest that as many as 14% of children aged 0 to 5 have private insurance that does not cover at least 1 universally recommended vaccine.24 An increasing number of state school and child care entry immunization requirements for varicella vaccine25 may persuade parents to pay out of pocket for the vaccine if it is not covered in private plans or to make the extra effort to bring their children to clinical sites where doses purchased through federal and state immunization programs are available.
Implications for Future Effects of Varicella Vaccine
Our findings prompt questions about medical, public health, and analytic strategies regarding varicella in the years ahead as the epidemiology of the illness shifts to a progressively older population at generally higher risk for complications.
First, doubts about the persistence of immunity after vaccination against varicella have long concerned clinicians and public health officials26–28 and resurfaced with an outbreak of varicella in a child care setting in 2002.29,30 If clinical and community studies of long-term immunity from varicella vaccination lead to future recommendations for booster doses for previously immunized individuals, then it will be illustrative to examine associated patterns of severe illness before and after such new recommendations.
Second, as presented in the Recommended Adult Immunization Schedule issued by the Advisory Committee on Immunization Practices of the Centers for Disease Control and Prevention,31 varicella vaccine is currently recommended for catch-up doses in nonimmune adults with chronic conditions such as diabetes, heart disease, chronic pulmonary disease, chronic liver disease, end-stage renal disease, and asplenia. Children's health care providers can assist in the adult vaccination effort by educating parents about these indications for varicella vaccine.
Third, questions remain about the incidence of herpes zoster in the varicella vaccine era. As had previous investigators,5,6 we excluded hospital discharges with herpes zoster diagnosis codes to avoid misclassification of the discharge as primary varicella infection. Herpes zoster was not the focus of this analysis, but the NIS data set does afford investigators the opportunity to examine trends in herpes zoster hospitalizations and corresponding charges before and after the implementation of universal childhood varicella recommendations.
Fourth, even with as many as 7.5 million unweighted discharges annually in the NIS, the ability to derive weighted national estimates from such data will become progressively limited as the immunization program becomes more effective in preventing varicella-related hospitalizations. As the incidence rates of varicella decline further, state and community surveillance systems7 will take on paramount importance because of their measurement sensitivity, notwithstanding their more limited generalizability. Such surveillance systems are essential to efforts both to document progress toward disease eradication and to detect persistent pockets of underimmunization in communities that remain vulnerable to vaccine-preventable diseases such as varicella.
As with all analyses of discharge data, our findings must be interpreted with the caveat that there may have been misattribution or inadvertent exclusion of ICD-9-CM codes pertinent to the study. Nonetheless, our estimated annual rates of varicella-related hospitalization were similar to those reported by authors who used different data sets.5–7,32
In the years 1993 through 1997, we observed year-to-year fluctuations in varicella-related hospitalization patterns consistent with annual variation in the incidence of varicella disease that has been reported previously in the United States and Europe.6,7,33–35 Although such fluctuations raise the question of whether VRHD rate declines from 1997 through 2001 might conceivably represent a persistent decline in natural disease incidence unrelated to immunization efforts, the magnitude of the observed effect and the timing concordant with rising immunization rates argue otherwise.
It was not possible to determine from the NIS data whether individuals had received varicella vaccine, so our analysis cannot provide a measure of vaccine efficacy. In addition, it is possible that the VRHD patterns in the states that participated in the NIS during the study period were different from the states that did not participate. Of note, state participation in HCUP databases is not based whatsoever on states' immunization efforts.
As with all cross-sectional rather than longitudinal data, we cannot definitively draw a causal link between national varicella vaccination efforts and decreased hospitalization rates and costs for severe varicella-related disease. Rather, we have described a strongly suggestive, time-specific association. It is possible that the use of antiviral medications such as acyclovir for treatment of primary varicella disease36 during the same time period may have contributed to declines in hospitalization rates. We are not aware of any published utilization data for antiviral varicella therapy that would permit the study of such a relationship. Moreover, other authors have observed declines in the incidence of varicella illness—which would not be affected by antiviral use—that correspond to rising community immunization rates7 and have linked these illness rate declines to decreases in hospitalization rates.
Providers and parents have worked together to vaccinate children against varicella, and rates of hospitalization for varicella-related illnesses have declined significantly for young children in the age group targeted for vaccination and for older children and adults as well. As immunization rates rise and stabilize among the youngest children in the future, vaccination of susceptible older children and adults will likely take on even greater importance because of the increased risk of morbidity and mortality among those individuals.
Ultimately, future health successes and economic benefits attributed to the childhood varicella immunization program in the United States shall be judged not only for their magnitude beyond what has been achieved to date but also in the breadth of their effects across different age groups, payers, and communities.
- ↵American Academy of Pediatrics. Recommendations for the use of live attenuated varicella vaccine. Pediatrics.1995;95 :791– 796
- ↵Halloran ME, Cochi SL, Lieu TA, Wharton M, Fehrs L. Theoretical epidemiologic and morbidity effects of routine varicella immunization of preschool children in the United States. Am J Epidemiol.1994;140 :81– 104
- ↵Nationwide Inpatient Sample (NIS). Healthcare Cost and Utilization in Project (HCUP). Rockville, MD: Agency for Healthcare Research and Quality; July 2003. Available at: www.ahrq.gov/data/hcup/hcupnis.htm. Accessed January 2, 2004
- ↵US Census Bureau, Administrative and Customer Services Division, Statistical Compendia Branch. Statistical abstract of the United States. Available at: www.census.gov/statab/www/. Accessed January 2, 2004
- ↵US Department of Labor, Bureau of Labor Statistics. Consumer price index. Available at: ftp://ftp.bls.gov/pub/special.requests/cpi/cpiai.txt. Accessed January 2, 2004
- ↵Meyer PA, Seward JF, Jumaan AO, Wharton M. Varicella mortality: trends before vaccine licensure in the United States, 1970–1994. J Infect Dis.2000;182 :383– 390
- ↵Brunell PA. Chickenpox. In: Wehrle PF, Top FH, eds. Communicable and Infectious Diseases. 9th ed. St. Louis, MO: The CV Mosby Company; 1981;155–160
- Englund JA, Balfour HH Jr. Varicella and zoster infections. In: Hoeprich PD, Jordan MC, Ronald AR, eds. Infectious Diseases. 5th ed. Philadelphia, PA: JB Lippincott Company; 1994;952–962
- ↵Zaia JA, Grose C. Varicella and herpes zoster. In: Gorbach SL, Bartlett JG, Blacklow NR, eds. Infectious Diseases. 2nd ed. Philadelphia, PA: WB Saunders Company; 1994;1311–1323
- ↵Centers for Disease Control and Prevention. Estimated vaccination coverage with individual vaccines and selected vaccination series among children 19–35 months of age by state—Q1/1996–Q4/1996. Available at: www.cdc.gov/nip/coverage/NIS/96/toc-96.htm. Accessed January 2, 2004
- Centers for Disease Control and Prevention. Estimated vaccination coverage with individual vaccines and selected vaccination series among children 19–35 months of age by state—Q1/1997–Q4/1997. Available at: www.cdc.gov/nip/coverage/NIS/97/toc-97.htm. Accessed January 2, 2004
- Centers for Disease Control and Prevention. Estimated vaccination coverage with individual vaccines and selected vaccination series among children 19–35 months of age by state—Q1/1998–Q4/1998. Available at: www.cdc.gov/nip/coverage/NIS/98/toc-98.htm. Accessed January 2, 2004
- Centers for Disease Control and Prevention. Estimated vaccination coverage with individual vaccines and selected vaccination series among children 19–35 months of age by state—Q1/1999–Q4/1999. Available at: www.cdc.gov/nip/coverage/NIS/99/toc-99.htm. Accessed January 2, 2004
- Centers for Disease Control and Prevention. Estimated vaccination coverage with individual vaccines and selected vaccination series among children 19–35 months of age by state—Q1/2000–Q4/2000. Available at: www.cdc.gov/nip/coverage/NIS/00/toc-00.htm. Accessed January 2, 2004
- ↵Centers for Disease Control and Prevention. Estimated vaccination coverage with individual vaccines and selected vaccination series among children 19–35 months of age by state—Q1/2001–Q4/2001. Available at: www.cdc.gov/nip/coverage/NIS/01/toc-01.htm. Accessed January 2, 2004
- ↵Gold MR, Siegel JE, Russel LB, Weinstein MC. Cost-Effectiveness in Health and Medicine. New York, NY: Oxford University Press; 1996
- ↵Institute of Medicine. Calling the Shots: Immunization Finance Policies and Practices. Washington, DC: The National Academies Press; 2000
- ↵Institute of Medicine. Financing Vaccines in the 21st Century: Assuring Access and Availability. Washington, DC: The National Academies Press; 2003
- ↵Immunization Action Coalition. Varicella prevention mandates. Available at: www.immunize.org/laws/varicel.htm. Accessed January 2, 2004
- ↵Centers for Disease Control and Prevention. Recommended adult immunization schedule, 2003–2004. Available at: www.cdc.gov/nip/recs/adult-schedule.htm#print. Accessed January 2, 2004
- ↵Gordon JE. Chickenpox: an epidemiologic review. Am J Med Sci.1962;224 :362– 389
- Joseph CA, Noah ND. Epidemiology of chickenpox in England and Wales, 1967–1985. BMJ.1988;296 :673– 676
- ↵Arvin AM. Antiviral therapy for varicella and herpes zoster. Semi Pediatr Infect Dis.2002;13 :12– 21
- Copyright © 2004 by the American Academy of Pediatrics